AI’s 6 Percent Problem [Sponsored]
Tech.eu Tech.eu Editorial Team ● Covered by 7 sources
AI is everywhere in companies, but only 6% say it really moves profit. McKinsey says the missing piece isn’t the model — it’s ownership and workflow.
Based on reporting by Tech.eu, Tech.eu Editorial Team — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Companies have finally stopped treating AI like a lab curiosity. Nearly 90 percent now use it in at least one business function, and 44 percent say they’ve scaled it across the organization. But McKinsey’s latest State of AI, released in late August, says the money is still not following the hype. Only 37 percent report a measurable impact on EBIT, and that figure has barely moved from last year.
The real outlier is tiny. About 6 percent of companies count as true high performers: they say AI contributes at least 5 percent of EBIT, and they see that contribution as significant. That gap is not being closed by chasing the next model. It’s being closed, or missed, by the less glamorous stuff — clean data, clear ownership, and workflows that actually change.
The source of the pain is familiar to anyone who has watched an AI project get stuck in corporate mud. Customer data sits in one system, inventory in another, revenue in a spreadsheet on someone’s laptop. Ask an AI tool whether something is in stock, and it can get two different answers from two systems before confidently picking one. That is how companies end up with the old problem wearing new clothes: errors that human beings still have to fix afterward.
Ownership is the other trap. IT gets the tech, the business side owns the use case, management wants results, and nobody owns the whole thing. Then the pilot ends, and it gets filed away with the other experiments nobody wants to admit failed. The companies seeing returns do something more disciplined. They fix the data foundation first, put AI inside the tools people already use, and give one person responsibility for the outcome.
And they don’t spray pilots everywhere. Nearly three-quarters of the high performers have redesigned workflows around AI instead of just layering AI on top of the old process. A year earlier, that share was 55 percent. They also judge success where it counts: on the P&L, not in vague talk about hours saved. That theme showed up again at DMEXCO 2026 in Cologne, where more than 40,000 attendees and around 1,000 speakers kept circling the same question: what does it take for AI to work in the real world?
My take — AI-written commentary, not fact-checked reporting
The market has spent two years pretending model quality is the main story. It isn’t. The boring stuff wins: data plumbing, workflow design, and one human who actually owns the mess when the demo ends. The rest is just expensive theatre with better branding.
Read more about this at: Tech.eu
